3754
Biochemistry 1980, 19, 3754-3765
Patterson, A., & Ettinger, R. (1960) 2. Elektrochem. 6 4 , 98. Pople, J. A., Schneider, H. J., & Bernstein, W. G. (1959) High-Resolution Nuclear Magnetic Resonance, McGrawHill, New York. Ribas Prado, F., Giessner-Prettre, C., Pullman, B., & Dandey, J.-P. (1979) J . Am. Chem. SOC.101, 1737. Roberts, G. C. K., Feeney, J., Birdsall, B., Kimber, B. J., Griffiths, D. V., King, R. W., & Burgen, A. S. V. (1977)
in N M R in Biology (Dwek, R. A., Campbell, I. D., Richards, R. E., & Williams, R. J. P., Eds.) p 95, Academic Press, London. Sarma, R. H., & Mynott, R. J . (1972) Org. Magn. Reson. 4 , 577. Stone, D., & Smith, S. (1979) J . Biol. Chem. 254, 10857. Stone, D., Phillips, A. W., & Burchall, J. J. (1977) Eur. J . Biochem. 7 2 , 613.
Interaction of Bovine Carbonic Anhydrase with (Neutral) Aniline, Phenol, and Methanol? Gary S. Jacob,t Rodney D. Brown, 111, and Seymour H. Koenig*
ABSTRACT: We have investigated the interaction of bovine carbonic anhydrase with neutral aniline, phenol, and methanol molecules. The measurements are of optical spectra and solvent water and methanol proton magnetic relaxation rates of solutions of Co*+-subStitutedenzyme. We recently proposed a model [Koenig, S. H., Brown, R. D., & Jacob, G. S. (1980) Proceedings of the Symposium on Biophysics and Physiology of Carbon Dioxide, Springer-Verlag, West Berlin and Heidelberg], based on the interaction of enzyme with monovalent anions, that accounts for the pH dependences observed for a wide variety of phenomena, including the apparent pKa for enzymatic activity. We now extend the model to include the observed effects of neutral molecules. Aniline and phenol, though isoelectronic, shift the observed pKa values in opposite directions, and both appear to bind at the aromatic binding site to which sulfonamide inhibitors and aromatic esters are known to bind. The resulting binary complexes behave as altered enzymes, with different values of the pKa for activity, but otherwise are similar to the native enzyme. In terms of
our model, aniline and phenol alter the relative affinities of water and anions for the same coordination position of the metal ion at the active site. The effect is opposite in sign for the two molecules because of the differing proton affinities of the N H 2 and O H moieties of the phenol ring in each case. By extension, our results indicate that data from experiments using aromatic buffers such as imidazole and lutidine should be analyzed with some care; effects previously attributed to buffer in solution may well be due to binding of neutral buffer molecules to the aromatic binding site in the active region of the enzyme. The interaction of methanol with carbonic anhydrase is quite different, and very weak. Methanol does displace water at the metal, but to first order there is little, if any, preferential binding of methanol compared to water. Observations by others that alcohols inhibit esterase activity with inhibition constants on the order of 1 M are not attributable to binding of alcohol to enzyme but rather, in our view, result from the increased solubility of aromatic ester substrates in the alcohol-modified solvent.
%e zinc-containing enzyme carbonic anhydrase (EC 4.2.1 .l), which catalyzes the reversible reaction COz H 2 0 ~i H + HC03-, has also been found to catalyze the hydrolysis of a variety of esters (Tashian et al., 1964; Malmstrdm et al., 1964; Pocker & Stone, 1965, 1967) and the hydration of aliphatic aldehydes (Pocker & Meany, 1965a,b). A longstanding question has existed regarding the identity of the metal-activated ligand in the enzyme that is responsible for the nucleophilic attack on a bound COz during the hydration reaction. A widely held model for enzymatic activity, originally proposed by Davis (1959), identifies this nucleophile as a zinc-bound hydroxide ion. The observed pH dependence of the catalysis (greater hydration and esterase activity at higher values of pH) is then attributed to ionization of a metal-coordinated water ligand to produce the zinc-bound hydroxide nucleophile. This model, as well as most others proposed for the activity-controlling ionizing group (cf. Pocker & Sarkanen, 1978), cannot explain the observed rate of magnetic relaxation
of solvent water protons first reported by Fabry et al. (1970) (cf. Koenig & Brown, 1972). Moreover, Lindskog, a long-time proponent of the “hydroxide” mechanism, concludes a recent review by noting that “... as long as unequivocal evidence for the existence of Zn2+-bound OH- in the enzyme is lacking, this model must be continually questioned and tested against alternative models” (Lindskog, 1980). An alternate explanation for the pH dependence of enzymatic activity was recently proposed by us (Koenig et al., 1980). In our model, there is no ionization on the enzyme that determines the observed pH-dependent activity; rather, the pH-dependent phenomenon can be explained by the existence of both active enzyme and anion-inhibited enzyme, with relative concentrations that depend on the type and concentration of monovalent anion present in solution, and on pH. The active enzyme was proposed to have a zinc-bound water ligand, capable of exchanging rapidly with solvent, that can be replaced by a monovalent anion and an associated proton from solution to produce inactive enzyme. Thus, active enzyme predominates at high pH, and anion-inhibited enzyme predominates at low pH, with the pKa for activity being determined by the composition of the solvent. The essence of the model is that a constant charge environment is maintained in the active site, both statically and
+
+
‘From the IBM Thomas J. Watson Research Center, Yorktown Heights, New York 10598. Receiued February I, 1980. A preliminary report of this work was presented at the 1lth International Congress of Biochemistry, Toronto, Canada, July 1979. *Present address: Monsanto Corp., Corporate Research Laboratory, St. Louis, MO 63166.
0006-2960/80/0419-3754$01.OO/O
0 1980 American Chemical Society
N E U T R A L UOLECULES
\UD CARBO\IC
.\\HYDRASE
\OL.
19.
YO.
16, 1980
3755
dynamically. We proposed that a molecule can bind a t the active site only if the charge is not altered. In those preparations in which monovalent anions are excluded, it appears that HS04- anions, in equilibrium with the SO,,- generally present to maintain ionic strength, inhibit the enzyme at low pH. To test these ideas further, as well as to generalize their applicability to a wider range of phenomena, we have investigated in detail the interaction of carbonic anhydrase with small neutral molecules. We chose three compounds: aniline and phenol, aromatic and isoelectronic, which can interact with the aromatic binding site of the enzyme, and methanol, a weak inhibitor, a t least of esterase activity, that presumably mimics H 2 0 in its mode of binding. Very little is known about the mechanisms by which neutral molecules interact with carbonic anhydrase. Indeed, an extensive study of both hydration of C 0 2and hydrolysis of esters in the presence of neutral molecules such as methanol and phenol has not yet been reported. The inhibitory effect of neutral molecules on the esterase activity of the enzyme was first reported by Verpoorte et al. (1967) for the human enzymes, and investigated by Pocker & Stone (1968) for the bovine enzyme. More recently, Appleton & Sarkar (1975) observed that the inhibition constant for methanol decreased with increasing pH; that is, the apparent affinity of the enzyme for methanol increases with increasing pH. In addition, they reported a pK, for inhibition by methanol of 6.4, the same as the pK, for enzymatic activity. Both shifted to higher pH in the presence of increasing concentrations of monovalent anions. Methanol appeared to bind to enzyme in a one to one stoichiometry, competitively with monovalent anions. Aniline, too, was found to interact a t a single binding site but, unlike methanol, has a greater affinity for enzyme a t low pH, much like monovalent anions. However, the question of whether the neutral molecules interact by binding directly to the metal was left unanswered. Recently, Westerik et al. (1978) reported additional results for the interaction of methanol and aniline with carbonic anhydrase. Aniline was reported to bind directly to the metal of Co2+-substituted enzyme, yet did not effect the water proton magnetic relaxation rate. They proposed that aniline either binds at a fifth coordination site or. alternatively, displaces one of the histidine ligands of the metal. On the basis of ')C relaxation rate studies, it was claimed that methanol binds at a hydrophobic region of the enzyme -6 A from the metal. In the experiments reported below, we used Co2+-substituted bovine carbonic anhydrase in order to monitor changes in the active-site region. The experiments were designed to demonstrate, first, that the pH-dependent properties of the samples depend on anion concentration in the expected manner, and then to measure the alteration of these properties in the presence of aniline, phenol, or methanol. W e conclude that aniline and phenol bind near, but not to, the active metal to produce a binary complex that is essentially an altered, but otherwise active (with regard to hydration and dehydration), enzyme. The functional group of the inhibitor has the effect of altering the relative affinities of water and monovalent anions (with their associated protons) for the metal. By contrast, we find that methanol competes directly with H,O for a ligand of the metal. Our results will be compared with those of Westerik et al. (1978), and a number of apparent contradictions between the conclusions of the two reports will be resolved.
et al., 1970) to yield the B enzyme (akin to the human C), was demetallized in the usual way (Lindskog & MalmstrBm, 1962) and lyophilized after extensive dialysis against distilled water (Bertini et al., 1978). Co*'-substituted enzyme in deionized water (Co*+-BCA)' was prepared by dialysis of apoprotein against 1 m M CoC1, for 1 day and then against deionized water, with frequent changes, for 6 days at 5 "C. (The small amount of C1- anion introduced with the metal has negligible inhibiting effect, since the affinity of C1- for the active enzyme is relatively low.) The final pH, 5.4-5.7, is near the isoelectric point of the enzyme. Sample pH was measured with a Radiometer PHM65 meter and combination electrode from Microelectrodes, Inc., Ultra-high purity sodium chloride and sodium sulfate were obtained from J . T. Baker and E M Laboratories, respectively. Deuterated methyl alcohol (CD,OD) and methyl alcohol-d (CH,OD) were from Sigma and Merck, respectively. All other chemicals were reagent grade. Absorbance Measurements. Optical absorption spectra were measured at room temperature in cuvettes of I-cm path length, using a Cary Model 14 recording spectrophotometer equipped with 0 to 0.1 and 0 to 1 optical density slide-wires. Protein concentration was computed using Alcml%= 18 at 280 nm. All glassware used in the titration experiments was immersed in 0.1 M EDTA solution for at least 1 day and rinsed with deionized water before use. The p H dependence of the optical properties of the exhaustively dialyzed enzyme in the presence of chloride was measured by adding chloride as a concentrated salt solution to the dialyzed sample to give the desired anionic concentration. The p H was then adjusted with HCI, and a series of optical measurements taken as the pH was raised by the addition of N a O H . The optical titration of Co2+-BCA in the presence of aniline was done in a similar fashion. The neutral molecule was added to give a concentration of 20 m M , and a series of optical spectra taken as the p H was raised with K a O H ; the same sample was subsequently used for titrations in the presence of chloride and aniline. Chloride was again added as a concentrated salt solution; the p H was lowered with HCI to a minimum p H starting point and a titration carried out by raising the p H with NaOH. Titrations were performed at 0.06 and 0.6 M chloride in the presence of 20 m M aniline. Titrations of Co*+-BCA in the presence of phenol or methanol were performed in a fashion similar to the aniline titration. Final concentrations of the individual components are given in the legends to the figures. For the experiments using C H 3 0 H / D 2 0 , pD values are the uncorrected meter readings. The dilution of any sample at the end of a titration was always less than 1%. The fraction of active enzyme at each pH, here defined as the fraction of enzyme with the high-pH optical spectra, was calculated from the amplitude of the 640-nm peak relative to the initial plateau observed at high pH. This procedure is somewhat arbitrary, since the optical absorption does not remain perfectly pH independent at values of pH well beyond the upper (high pH) end of a particular titration. The procedure generally used was first to make a rough estimate of the pK, of a particular titration curve, then to take the amplitude of the absorption either 1.5 or 2 pH units above the estimated pK, as being a t either 97 or 99% of the titration limit, respectively. The data were then fit to a single
Experimental Procedures Materials. Native bovine anhydrase, purchased from Sigma and purified by chromatography on DEAE-cellulose (Kandel
* The KC1 flow rate from the reference electrode is guaranteed to be
' Abbreviation used:
BCA, bovine carbonic anhydrase.
no greater than 0.01 pL per 24 h, thus obviating the possibility of contamination of the sample by chloride.
3756
JACOB, BROWN, AND KOENIG
BIOCHEMISTRY 1.2,
I
I ,
,
1 7 1 T
W
IO-
tE
1
080604-
L
